[jrehor]

I don't understand this article at all. The pension fund uses block trades. Who takes the other side of those trades and how are they compensated?

I suspect that an equity desk takes the block, parcels it out into lots of small pieces, and works the market to get it off their books. Of course, they charge for that service, both in commissions and spread. So effectively, instead of paying HFT firms for providing liquidity, they're paying an investment bank equity desk. Does this really save money? If it does, why all the hoopla about HFT if you can avoid them by going through an equity desk?

Don't tell me that they just put their block on IEX and the tooth fairy fills it without price impact. That would be some serious magic.

[wdewind]

I agree with you, and I think the author, like most people who are writing against the "flash boys," fundamentally misunderstands what's going on.

I think the "black and white" rule that the large firms who feel they are being hurt by HFT want is the ability to execute a full trade before it impacts the market (since otherwise it's impossible to describe the exact moment they feel their trade is a signal to be traded off of). I think that's highly unreasonable, but if you do believe that I think IEX is probably relatively successful at it. It's totally possible that a block could be fully sold off on IEX before the 350ms latency allows that information to escape to the outside world and be traded against, no? Market impact is ultimately about the same, just delayed enough to allow the seller to escape.

I would happily be correct on any of this, since I don't work in finance.

[jrehor]

It is possible that a block would be sold within 350ms, but how likely is it? That's a function of block size. If you have 100 shares, you'll get it done 100% of the time. 1,000 shares? Somewhat less than 100%. 100,000 shares? Rarely.

These guys are ginormous. $100m block is about a 0.01% position for them. You can't just put a limit order on IEX (or any other exchange for that matter) when you're dealing with this magnitude. You need a team of specialists working full time disguising the order and parceling it out to multiple venues, and it will usually take days to get it done.

The other way of moving a block this size is by finding a natural counterparty and negotiating a deal directly with them. The trade will then get booked through a broker-dealer for reporting and settlement purposes. Those kinds of negotiations take days, too, and you can't avoid showing your hand. Plus, it's not a very scalable solution.

[lrm242]

IEX only delays the order entry side of the system. Market data (they are dark atm so no quotes/book feed of course) and trade reporting is not delayed. The magic shoe box is there to stop a very specific situation from occuring: IEX pricing pegged orders against stale market data.

[wdewind]

So I don't fully understand what IEX does then. Let's say A puts 500 shares of AAPL onto IEX at $100/share, and B puts a buy order in for 500 shares at $100/share, but in between posting the BUY and the order executing (in that 350ms), they get more information, and decide they actually want to pull out, are they unable to?

Let's say B instead only buys 250 shares, what happens to the remaining 250 shares?

[lrm242]

They are unable to do so. It is a queued system with a fixed transit latency through the queue of 350us. Any remaining un-executed of A is left on the IEX book.

IEX has this transit latency because they have order types that offer pegged execution, i.e. midpoints. A midpoint order is executed at the midpoint of the prevailing NBBO. The data required to build a view of the prevailing NBBO comes in from all other lit venues.

IEX has a 350us delay because they want to slow down the order entry side to give themselves enough time to make sure they are not pricing their pegged orders using stale market data. Simple as that.

[lrm242]

It's 350us btw. Micros.... not millis.

[lrm242]

Sell side traders or block crossing networks (where the other side is taken by another block trader).

Sell siders are compensated because they will build into the price the expected impact. The pension fund gets a known price impact with essentially 0 variance and the sell sider takes on the risk of execution to offset the trade, and essentially earn a spread between what the pension fund paid them for the risk transfer and their 'skill' in getting the trade done in the market.

It obviously can and will be more complicated than that with varying benchmarks and compensation schemes, but that is the jist of it.

[jim_greco]

Very odd indeed. This is a complete regression to how the market was traded 20 years ago when every bank used to have a NASDAQ market making desk that would make a principal market in any security you wanted. Those desks all got wiped out when spreads collapsed because HFT firms could do it for a fraction of the cost.

[lrm242]

They had a big hand in wiping themselves out when they got caught colluding.